影像图中水体识别与提取技术研究综述

Review of River Recognition and Extraction Techniques in Images

随着工业的发展与生活水平的提高,水资源的需求逐渐升高,在水资源的高效动态化管理等方面提出了更高的要求。科学监测水量、水质及水体分布有助于了解各地水资源的流域分布、水体变动和污染情况。目前,水体的大范围监测主要依靠人工野外勘测,耗时、危险、无法实时监控水体动态,卫星遥感数据可以高效、准确地获取水体各类形态参数,涉及位置、面积、河宽等多个信息。随着遥感卫星测绘数据量的增加、空间分辨率的提升以及影像识别算法的深入研究,智能化高效解释水体信息是一项重要的研究课题。基于合成孔径图像与光学图像,总结影像图识别算法通用性与匹配校准,对高精度全自动化影像信息提取算法进行了展望。

With the development of industry and the improvement of living standards,the demand for water resources is gradually increasing,and higher requirements are put forward in the efficient dynamic management of water resources.Scientific monitoring of water quantity,water quality and water distribution is helpful to understand the watershed distribution,water changes and pollution of water resources in different regions.At present,large-scale water body monitoring mainly relies on artificial field survey,which is time-consuming,dangerous and unable to monitor water body dynamics in real time.Satellite remote sensing data can obtain various water body morphological parameters efficiently and accurately,involving multiple information such as location,area and river width.With the increase of remote sensing satellite mapping data,the improvement of spatial resolution and the in-depth study of image recognition algorithm,the intelligent and efficient interpretation of water information is an important research topic.Based on synthetic aperture image and optical image,the generality and matching calibration of image recognition algorithm are summarized,and the high-precision fully automatic image information extraction algorithm is prospected.